A two-person effort: On the role of the agent in EDA-DMILS experiments - electrodermal activity - Direct Mental Interaction With Living Systems - Statistical Data Included

**********

For over two decades, parapsychologists have accumulated empirical evidence suggesting a so-called Direct Mental Interaction With Living Systems (DMILS). This unorthodox form of interaction involves changes in one person's autonomic nervous system covarying with a second, distant person's intentions (see Schlitz & Braud, 1997). From the numerous parameters of the autonomic nervous system, electrodermal activity (EDA) is by far the most favored indicator for the DMILS effect: Being largely affected by unconscious processes (Boucsein, 1995; Edelberg, 1972; Roy, Sequeira, & Delerm, 1993), inferences drawn from EDA allow a more reliable interpretation than, for instance, inferences drawn from physiological signals, which are more susceptible to arbitrary (intentional) control (e.g., heart rate).

To date, there is no commonly accepted model of how this distant mental interaction operates. There seems to be a consensus, however, that specific favorable states of the two systems (agent and receiver) are conducive to the effect (see Braud, l980a, 1980b). At the beginning of experimental DMILS work, Braud postulated that two systems best "fit" or "conform" when one, the target system, is relatively labile and the other, the influencing system, is relatively stabile. In line with this view, Braud and Schlitz (1983) found lability, as operationalized by more FDA at rest, to be a potential prerequisite for the agent's calming effort. Unfortunately, there are no systematic follow-up studies on the significance of such state variables. In a recently conducted series of DMILS experiments, electrodermal lability was not shown to be an important variable for the DMILS effect (Schneider, in press).

However, the DMILS effect is usually regarded as the result of a joint endeavor (Braud & Schlitz, 1991), and the experimental setting is sometimes referred to as a healing analogue (Schlitz & Braud, 1985, 1997). Hence, there have been various attempts to identify important factors on behalf of both agent and receiver. Among these are, for instance, the participants' relatedness (Delanoy, Morris, Brady, & Roe, 1999), experimenter effects (Delanoy & Morris, 1999), and blocking or cooperating strategies (Braud, Schlitz, Collins, & Klitch, 1984; Watt, Ravenscroft, & McDermott, 1999). Unfortunately, the search for moderating variables has not been very successful in explaining how the effect is brought about. As a consequence, the process-oriented approach to DMILS rather resembles the figurative search for the needle in the haystack: No firm conclusions can be drawn as to what variables are to be considered for a successful DMILS study.

So far, only little attention has been devoted to the actual necessity of an agent and, therefore, the role of the agent is underresearched. In the DMILS paradigm, no studies have addressed this topic. A remote staring study (an experimental setting very similar to DMILS) by Braud, Shafer, and Andrews (1993) found that an experimental control condition ("sham"), in which the data were recorded without any starer present, produced no significant differences in the starees' EDA.

Results from other parapsychological paradigms involving two interacting individuals have not produced any significant conclusions either. For example, surveys of the ganzfeld literature by Palmer (1978) yielded rather heterogeneous results. From the few studies applying a systematic comparison between the presence or absence of an agent and the experimental outcome, some attributed only little importance to the agent's presence (e.g., Carpenter, 1977). Other studies yielded better results with an agent present (Raburn & Manning, 1977). Still others found no differences between different sending conditions (Morris, Dalton, Delanoy, & Watt, 1995; Williams, Roe, Upchurch, & Lawrence, 1994).

From a theoretical point of view, there are numerous ways how the DMILS effect might occur without any agent being present at all (Stokes, 1987). For example, one could conceive of the effect as the consequence of the receiver's precognition. Thus, a person would "foresee" the sequence of influencing epochs (activating and calming) and adjust his or her physiological arousal accordingly. Obviously, this line of reasoning would not regard the agent as a necessary prerequisite (although he or she might have a relevant psychological impact). However, if an intentionally active agent is crucial for the DMILS effect, it might be worthwhile to explore his or her role in more detail.

This apparent lack of consistent empirical evidence led us to design a study in which the agent's role in the DMILS setting was systematically explored. We compared sessions with and without an agent being intentionally active (i.e., attempting to activate or calm the receiver). Because there are various ways to assess EDA (Boucsein, 1992) and because little is known about the functional significance of various EDA parameters (which parameter best indicates an effect?), this study was also designed to explore these issues in more detail. This topic is vital for the DMILS paradigm because the majority of the published DMILS studies do not use standard psychophysiological methodology (Schmidt & Walach, 2000) and, therefore, the size of the true DMILS effect is unknown (Schneider, Binder, & Walach, 2000). Hence, it is questionable whether the reported effect by Schlitz and Braud (1997) is subject to artificial biases. For example, in a recently conducted EDA-DMILS pilot study by Schmidt, Schneider, Binder, Burkl e, and Walach (2001), various EDA parameters yielded comparable effect sizes clearly larger than the one reported in Schlitz and Braud's meta-analysis (ES= .4 to .46 vs. ES= .25; with ES representing Rosenthal's r). Yet, in another four DMILS studies (Schneider, in press), only one EDA parameter (number of nonspecific skin conductance responses [NS.SCR freq]) turned out to be a functional indicator with significant effect sizes of -.49 (Experiment 3) and .52 (Experiment 4). Hence, it is unclear which EDA parameters should be included to best reflect a DMILS effect.